THE DETERMINATION OF WATER SURFACE MORPHOLOGY AT RIVER CHANNEL CONFLUENCES USING
AUTOMATED DIGITAL PHOTOGRAMMETRY
AND THEIR CONSEQUENT USE IN NUMERICAL FLOW MODELLING
J.H. Chandler,
Department of Civil Engineering,
Loughborough University,
Loughborough, Leicestershire LE11 3TU, UK.
EMail: J.H.Chandler(g]Lboro.ac.uk
S.N. Lane and K.S. Richards
Department of Geography,
University of Cambridge,
Downing Place, Cambridge CB2 3EW, UK.
Commission VII, Working Group 5 Terrestrial Ecosystem Monitoring
KEY WORDS: hydrology, close-range, non-metric, DEM/DTM, automation, surface reconstruction, geomorphology, low-cost.
ABSTRACT:
This paper describes the development and application of automated digital photogrammetry to derive the 3D coordinates of a
dynamic and turbulent water surface of an actively braiding pro-glacial stream in the Swiss Alps. A net of surface marker targets
was constructed using cheap polystyrene balls constrained by a series of fine lines. Stereo imagery was acquired using a pair of
synchronised semi-metric Hasselblad cameras and scanned at a resolution of 20 microns. Image coordinates were measured
automatically using Visilog, a general image processing package, transformed into photo-coordinates and sorted automatically
using the collinearity condition. Final object coordinates were derived using a self-calibrating bundle adjustment, elevations
corrected for combined spherical offset and buoyancy. These surface morphological data are being used to assist the development
of a 3D computerised flow model.
elevation varies in both the cross-stream and downstream
. i ions. T ura surement of water surface data ha:
1. Introduction directions. The accurate measurement of water surface data has
been shown to be critical to understanding flow processes in
natural rivers (e.g. Dietrich and Smith, 1984; Dietrich and
Whiting, 1989). Its measurement has previously been achieved
through the use of mechanical methods, normally lowering a
pin of known initial elevation onto the water surface from a
levelled and positioned base (Dietrich and Smith, 1984).
Whilst acceptable for small channels, providing the requisite
infrastructure is available, this is not the case for larger
channels, or situations where the channels are more dvnamic.
Water surface measurement in such contexts requires remote,
rapid and non-contact methods. This paper will describe the
experimental work developed to derive the true water surface
and is based upon automated digital photogrammetric methods.
Research into flow processes and sediment transport in natural
river channels is focusing upon detailed understanding of river
channel confluences. The flow structures created within such
regions are of particular relevance to environmental engineers
A where understanding the way fluid mixes is critical to
determining pollution dispersal processes. Numerically
simulating such processes using sophisticated computer models
is an important and productive method of gaining
understanding and insight into the complex interrelationships
between flow processes, sediment transport and channel form.
Rapid development of computer hardware has allowed these
simulated flow models to solve analytically, and in three
dimensions, ever more complex equations and relationships.
Such increased sophistication demands more accurate
‘boundary condition’ information to represent parameters such
as: the shape of the channel; flow rate; bed roughness and the
three dimensional surface of the water.
A
The method was developed and tested on an actively braiding
gravel bed river immediately downstream of the Upper Arolla
Glacier in the Swiss Alps. This demanding environment
necessitated the development of a robust technique capable of
withstanding and operating in flow velocities as high as 2ms"!.
The technique involved the use of floating marker points to
locate the position of a point just above the water surface. Six
lines were used to interconnect eight markers points, each line
attached to a 2.0m length of plastic piping. The whole 'net' of
markers could be floated to important regions of the confluence
whilst acquiring photogrammetric imagery. Conventional
photo-control points were established and coordinates derived
using 3D spatial intersection surveying methods. Images of all
targeted points were acquired using a pair of Hasselblad
cameras situated on a raised platform adjacent to the stream
A combination of analytical photogrammetry and tacheometric
survey of the sub-surface stream bed has proved an effective
method of deriving three dimensional terrain models (DTMs)
of rapidly changing stream channels in previous work, (Lane et
al., 1994a). DTMs have been used to provide initial boundary
| conditions for a numerical flow model (Lane et al., 1995) and
to investigate the spatial sensitivity of flow parameters through
distributed factor perturbation, (Lane ef al., 1994b). A planar
water surface had to be assumed for these analyses which for
many river channel problems is unsatisfactory as water surface
99
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B7. Vienna 1996